Does Smoking Cannabis Cause Schizophrenia?

Blogging on Peer-Reviewed Research

A lot of people are talking about a new study showing a 40% increase risk of “psychosis”, which I first heard news of in this story, from the Daily Mail:

A single joint of cannabis raises the risk of schizophrenia by more than 40 percent, a disturbing study warns.

The Government-commissioned report has also found that taking the drug regularly more than doubles the risk of serious mental illness.

Overall, cannabis could be to blame for one in seven cases of schizophrenia and other life-shattering mental illness, the Lancet reports.

Something sounds a little off. Let’s see what this Lancet study says.

Ok, first we see it is a meta-analysis, which means one should be cautious as the quality of the study is highly dependent on their design and inclusion criteria. Their findings were as follows.

There was an increased risk of any psychotic outcome in individuals who had ever used cannabis (pooled adjusted odds ratio=1·41, 95% CI 1·20-1·65). Findings were consistent with a dose-response effect, with greater risk in people who used cannabis most frequently (2·09, 1·54-2·84). Results of analyses restricted to studies of more clinically relevant psychotic disorders were similar. Depression, suicidal thoughts, and anxiety outcomes were examined separately. Findings for these outcomes were less consistent, and fewer attempts were made to address non-causal explanations, than for psychosis. A substantial confounding effect was present for both psychotic and affective outcomes.

Woah! Reefer Madness might have been right? Now, we have a modest effect, but a statistically significant dose-response relationship between marijuana use and psychosis. This is very interesting. What was the design?

Studies were included if they were population-based longitudinal studies, or case-control studies nested within longitudinal designs. We excluded cohorts of people with mental illness or substance-use-related problems, studies of prison populations, and RCTs of cannabis for medical use

Ok, I’m still with ya here. Reasonable criteria, good, good. How did they define “psychosis”:

Diagnostic outcomes for psychosis included schizophrenia, schizophreniform, schizoaffective, or psychotic disorders, non-affective or affective psychoses, psychosis not otherwise specified, psychotic symptoms, delusions, hallucinations, or thought disorder. Presence of delusions, hallucinations, or thought disorder was a requirement for all psychosis outcomes. Affective, mood, or bipolar disorder, affective disorder not otherwise specified, depression, suicidal ideation or suicide attempts, anxiety, neurosis, and mania were included for affective outcomes.

Here a slight problem, which the authors acknowledge, is that intoxication effects or descriptions of intoxication effects could lead to false positive psychosis diagnosis. But that’s ok, science is tough, you do what you can.

Their search criteria were good etc., so what were the results exactly? Well this one figure is probably the most important:

What this shows is the odds ratios from studies of psychosis in marijuana users for those that have ever smoked marijuana, with confidence intervals, compiled into this meta-analysis. The dose response one is similar but shows a bigger effect. Overall the odds ratios of these studies is 1.41 with a 95% CI of 1.20-1.65 which you can view as upper and lower limits of the probable size of this effect. Stunning no? Should we be convinced this is a real effect?

First of all, the statement that “just one joint” increases risk by 41% is absurd. The study here is of those who have tried marijuana once or more, not of people who have only tried it once. So already, the Daily Mail and every other news organization is way off. Second, I think we’re ultimately seeing a post-hoc ergo propter hoc argument, and a dose-response that’s more characteristic of the population studied than a real pharmacologic effect. Now, why do I think this is bogus?

Well for one, people with serious mental illness often do drugs. Lots of drugs. And not just marijuana. Marijuana is actually pretty low down on the list. In other words, people with schizophrenia and other serious mental health problems are at high risk of addiction. And this will necessarily confound their study.

The authors assert that this is not the problem, that is, mental health susceptibility is not the cause of the marijuana use but rather, marijuana use is increasing the probability of psychosis. Their reasoning is that the cohorts excluded populations that already had a diagnosis of schizophrenia or other psychotic disorder, so they are studying people who showed no signs of mental illness, then became mentally ill. This is well and good, and absolutely the right thing to do, otherwise you would immediately see a huge effect simply because people with serious mental health problems are at high risk of addiction. But, is the assumption still valid?

Well, no. And my answer lies in smoking. Yes, plain old cigarettes. Do you have any idea how much schizophrenics smoke? Tons. It’s unbelievable. In fact, approximately 90% of schizophrenics smoke, and of schizophrenics who smoke, 90% of them started smoking before onset of their illness. Further, evidence of a dose response effect, 68% of schizophrenics are heavy smokers, compared to 11% of the smoking population.

Now this is kind of astounding. Why don’t we say that cigarettes are putting people at risk for psychosis? Just because something follows an event, doesn’t mean that event is the cause. The population is choosing the drug, the drug is not creating the population. If you were to do similar study with cigarettes, or booze, you’d find a similar correlation (RR=1.94), and dose response etc (although some articles have suggested a protective effect as well – albeit in older cohorts). Eliminating the cohort that already had the diagnosis would not eliminate the effect, because schizophrenics overwhelmingly start smoking, and other drugs, before the psychotic break that defines the diagnosis.

Similar findings are often shown with alcoholism and early drinking, but suffer from the same assumptions. There are people with a tendency towards addiction, for various reasons, and at earlier ages they engage in drug-seeking behaviors. The mistake is thinking that when they become alcoholic or drug dependent in early life that the drugs caused the behavior, rather than considering the possibility that such behavior existed even before exposure to the drugs. Hell, I knew people growing up that would seek a high from every chemical at hand (long before they had even had access to pot or booze), not because they were addicted to those chemicals, or had even tried them before, but because they were the type that actively seek a high.

Correlation is not causation people. And this particular correlation is a hairy one, due to the tendency towards drug abuse that is characteristic of those suffering from psychotic disorders. I would put little trust in this particular study’s conclusions as the more likely phenomenon being studied is the predilection for drug-seeking in adolescents who subsequently become schizophrenic.

1. Theresa HM Moore, Stanley Zammit, Anne Lingford-Hughes, Thomas RE Barnes, Peter B Jones, Margaret Burke and Glyn Lewis, Cannabis use and risk of psychotic or affective mental health outcomes: a systematic review, The LancetVolume 370, Issue 9584, , 28 July 2007-3 August 2007, Pages 319-328.